In New York City, diners will be able to have a meal inside a restaurant at the end of the month, something that hasn’t happened there since the coronavirus pandemic began. In Florida, bars are set to reopen Monday for the first time since late June.

One decision appears to be riskier than the other, according to an analysis of cellphone and coronavirus case data by The Washington Post.

States that have reopened bars experienced a doubling in the rate of coronavirus cases three weeks after the opening of doors, on average. The Post analysis — using data provided by SafeGraph, a company that aggregates cellphone location information — found a statistically significant national relationship between foot traffic to bars one week after they reopened and an increase in cases three weeks later.

The analysis of the cellphone data suggests there is not as strong a relationship between the reopening of restaurants and a rise in cases, nor with bar foot traffic and cases over time, except for a handful of states.

But like with so much in the pandemic, easy answers can prove elusive.

A study by the Centers for Disease Control and Prevention of nearly 300 adults who tested positive for the coronavirus found that they were more than twice as likely to have dined at a restaurant in the two weeks before getting sick than people who were uninfected. Those who tested positive and did not have close contact with anyone sick were also more likely to report going to a bar or coffee shop. The same effect was not seen in visits to salons, gyms and houses of worship, or in the use of public transportation.

“You’re sitting there for a long time, everyone’s talking,” said Linsey Marr, an environmental engineer at Virginia Tech. “And that’s just a recipe for spread.”

Few states make their contact-tracing data available, but in two that do — Colorado and Louisiana — bars and restaurants are responsible for about 20 percent of cases traced to a known source. San Diego traced nearly one-third of community outbreaks to restaurants and bars, more than any other setting.

But Louisiana’s experience suggests bar patrons contribute more to the spread of the virus than restaurant diners. There have been 41 outbreaks tied to restaurants and the same number of outbreaks associated with bars, but bar outbreaks appeared to result in more infections, with 480 cases traced to those establishments compared with 180 from restaurants.

Marr said indoor dining can be reasonably safe in a restaurant operating at 25 percent capacity and with a ventilation system that fully recirculates air every 10 minutes. New York City’s policy will allow for only 25 percent capacity at first, with a scheduled increase to 50 percent in November if transmission rates remain low.

Still, Marr said, she “will not eat inside a restaurant until the pandemic is over.” As one of the first scientists to begin emphasizing that the virus was spread primarily by air, she has been concerned about indoor drinking and dining since March.

“People go to restaurants to talk,” she said, “and we know that it’s talking that produces a lot — 10 to 100 times more — aerosols than just sitting.”

Other countries facing outbreaks imposed stricter and longer shutdowns on bars and restaurants. Ireland has yet to open its pubs. Countries that did reopen bars and restaurants have, like American states, scaled back in the face of fresh outbreaks.

Researchers at the Massachusetts Institute of Technology say that because U.S. policies vary by state and county, waves of closures and reopenings may have perversely led to more viral spread, as people traveled to enjoy freedoms not allowed closer to home.
The National Restaurant Association argues that restaurants are safe if they follow proper mitigation guidelines and that the industry has been unfairly maligned by the actions of an irresponsible few.

“Bars become particularly risky,” said Larry Lynch, who handles food science for the restaurant trade group. “Anybody who had been in bars knows that conversations get louder, people get closer.”

But, he said, “we haven’t seen … any kind of systemic outbreaks from people going into a restaurant that’s practicing what we ask them to practice.”

Lynch questioned the methodology of the CDC study, noting it covered only 295 people and did not identify the sources of transmission.

The American Nightlife Association, which represents the bar and nightclub industry, did not respond to a request for comment.

Kristen Ehresmann, director of the Infectious Disease Epidemiology, Prevention and Control Division at the Minnesota Department of Health, agreed that when restaurants and bars abide by guidelines designed to reduce transmission, few cases of the coronavirus have been traced to those establishments.

But there are more than a few bad actors, she said: 1,592 cases of covid-19, the disease caused by the coronavirus, have been tied to 66 bars and restaurants in Minnesota. And in 58 other establishments, cases were reported among only staff members, resulting in 240 illnesses. One bar in St. Cloud, Minn., the Pickled Loon, was the only place visited by 73 people who got sick and was one stop among several for 44 other people.

“The bottom line is, we’re seeing a big chunk of our cases associated with these venues, and those cases go on to get other cases in other settings,” Ehresmann said. “We can’t ignore the impact.”

Iowa’s first big spike in coronavirus cases originated in the meatpacking industry. Then, says University of Iowa epidemiologist Jorge Salinas, came bars in college towns such as Iowa City, where he is based.

“It was very clear,” Salinas said. “We reviewed records for patients, and they all shared that common exposure of having been to a bar in the previous five days or so. Usually, the same bars that tend to be very crowded and very loud, rather than a place you just sit down to go and have a beer.”

He said those bars began closing not because of government intervention, but because so many staff members fell ill. By that point, the young people who got infected at bars had begun spreading the virus to an older population through family and work.

After about two months, the outbreak in Iowa City started to burn out. But then college students started returning to campus.

“It’s just a different group of young people but similar exposures — going to bars, hanging out, going to large parties,” he said.

He said an order from Gov. Kim Reynolds (R) closing bars in Iowa City and five other hard-hit counties was welcome but overdue. In the past two weeks, more than 1,000 young people in the region have become infected.

“Unfortunately, it’s late in the game,” Salinas said. “It would have been better if it had been done to prevent this rather than as a reaction to this.”

Politicians who favor an aggressive approach to containing the coronavirus have been hesitant to shut down bars and restaurants. Expanded federal unemployment benefits lapsed more than a month ago. Loans to small businesses are forgiven only if they can keep workers on the payroll, which is usually impossible while running at reduced capacity.

According to the Bureau of Labor Statistics, 2.5 million jobs in bars and restaurants have been lost since February. Although that’s an improvement from the spring, many restaurant owners say they are barely hanging on.

“Winter is coming, and I’m staring down the barrel of the gun of what’s going to happen,” said Ivy Mix, owner of a restaurant called Leyenda in New York and author of the book “Spirits of Latin America.” Even when indoor dining reopens, Mix said she is not sure she and her staff would be comfortable serving enough people inside her Brooklyn restaurant to make a profit.

“This is almost like being thrown a deflated life-jacket — the action and the symbolism is there, but the actual aid is not,” she said.

That’s why she says the only solution is federal legislation introduced by Rep. Earl Blumenauer (D-Ore.) that would issue $120 billion in grants to independent bars and restaurants. A Senate version would cover some chains as well.

“Eleven million people work for these independent restaurants,” Blumenauer said. “If we don’t do something, the evidence suggests that 85 percent of them are not going to survive this year.”

His office estimates that the legislation would more than pay for itself, generating $186 billion in tax revenue and unemployment savings. He said President Trump was receptive in a meeting with supporters earlier this year, but that in recent weeks the administration “has basically shut down meaningful negotiations.”

Arizona reopened indoor restaurants and some bars at the end of August, after a hasty spring reopening and more than 5,000 deaths.

“We really kind of reaped the whirlwind,” said David Engelthaler, a former state epidemiologist now at the Translational Genomics Research Institute. “A lot of that was driven by people going into bars and nightclubs, typically the 20- to 30-year-old set, interacting, socializing like they did prepandemic. And that just supported a kind of wildfire of cases.”

He said he thinks it is probably feasible to reopen restaurants at reduced capacity, but bars are a different story.

“One thing that all bars have in common is that they create a lowering of inhibition, and I think more than anything, this will cause the spread of covid,” he said. “We get more complacent, more comfortable, covid starts spreading.”

With temperatures still regularly topping 100 degrees in Arizona, the appeal of outdoor food and drink is limited. After a rapid May reopening led to a spike in cases and deaths, the state has just begun trying a more cautious approach.

Under Arizona’s new, more deliberate reopening, businesses must apply to reopen and bars must serve food to qualify. But Saskia Popescu, an epidemiologist based in Phoenix, said it is unclear whether those requirements are sufficiently stringent.

“If you put out two items does that count?” she asked. “I just worry that we’re kind of all doing this at once.” She noted that more than 500 new cases a day continue to be reported in Arizona, about the same as during the first reopening: “We’ll see if we learn from our lessons.”

Lenny Bronner, Jacqueline Dupree and Monica Ulmanu contributed to this report.

METHODOLOGY

Foot traffic data for this article was provided by SafeGraph, a company that aggregates location data from tens of millions of devices and compares it with building footprints. This doesn’t account for all devices — only those with one of the apps SafeGraph uses to track GPS location. Because the data relies on GPS pings and official building footprints, it doesn’t account for people who don’t have phones or those who are visiting outdoor parking lots.

Bar traffic in particular varies widely state to state, because until recently, SafeGraph’s data contained a relatively small sampling of bars, but also because the definition of a bar is somewhat loose. Visits data is normalized according to how many devices were reported each day.

Per capita calculations were based on coronavirus case counts tracked by The Post and data from the census. On average, states were seeing 3.7 cases per 10,000 residents the week they reopened. Three weeks later, that had increased to 6.9 cases per capita.

The correlation coefficient from comparing the increase in bar foot traffic and the increase in cases per capita three weeks later is .489 for 43 point. There is, of course, possible confounding by increases in other types of contact also coinciding with bar reopenings.